Control cohort (holdout)

A cohort deliberately excluded from every treatment, so it can serve as the untouched baseline lift is measured against.

A control cohort — often called a holdout — is a cohort that deliberately receives no treatment. It exists for exactly one reason: without a group that saw nothing, you can only measure attribution (how many people who saw a variant converted), not lift (how many more converted because they saw it, versus what would have happened anyway).

A holdout has to be a first-class cohort, not an afterthought bolted onto a report after the fact — a campaign that never designates one has no honest way to say what its treatments actually caused. See why every campaign needs a true holdout for the full argument, including the global-vs-per-campaign control tradeoff.